s1213555
Can state-of-the-art numerical modeling result in a more efficient and economic
design of power plants?
Supervisors:
Ir. Robin Morelissen, Deltares Dr. Ir. Pieter C. Roos, UT Dr. Kathelijne M. Wijnberg, UT
July 8, 2016
In coastal regions, some power plants use ambient water for cooling, which increases the temper- ature of the water used. The behavior of the outfall plume has an indirect effect on the capital and operational cost of a power plant. The design of a power plant configurations is most often based on computer model predictions. Currently, separate models are available for the near field and far field, which are both dependent on the input values and assumptions made by the mod- eler. This makes it difficult for non-expert clients to judge the quality of the model outcomes.
This study objectively investigated the added value of state-of-the-art recirculation modeling compared to a typical straightforward modeling approach for the optimization of the intake and outfall configuration of a power plant. Because of the wide range of variation in the above stated problem, the problem was assessed using a case study. This case was carefully chosen in order to capture the most important and relevant parameters such as the residual current, wind, ambient temperatures and a nearby river discharge.
Two models were set up, the first using a straightforward model approach (Approach A) and the second using state-of-the-art model approaches (Approach B). The aim of Approach A is to estimate the intake and outfall configuration with a quick and simple assessment. This was done by selecting a common consultancy practice model, a 3D far field model. The ambient conditions were selected based on common weather conditions. Approach B included state-of- the-art model approaches in order to assess the problems processes more physically correctly.
This was achieved extending Approach A with a dynamically coupled near-far field model and selecting the ambient conditions with the SBAM-method.
In order to assure an objective design process, a design framework was set up beforehand, which included 18 predefined designs and fixed criteria to select the ’best’ option. Based on the two models, two offshore intake and outfall configuration designs were proposed. The value of the two Approaches was evaluated base on offshore capital costs and recirculation costs.
In this case study, Approach A highly overestimated the temperature in the near field for all diffuser designs. Due to this, designs were rejected by our design framework that were found suitable in Approach B. The proposed design by Approach A will be located further into the sea resulting in a longer outfall pipeline. This results in an additional $1.035 million capital costs for the Approach A based design compared to the Approach B based design, an increase of 23%. Furthermore, additional maintenance can be tens of thousands of dollars per year and the operational costs will also be larger for a design with a longer pipeline system.
For all investigated designs, the yearly averaged intake temperature assessed by Approach A was within 30% of the assessed intake temperatures of Approach B. In terms of recirculation costs, this amounts to a difference of $300.000 in the lifetime of the power plant.
In conclusion, this case study helps to clarify that cases exist where an added-value for a state- of-the-art modeling approach can be found. In terms of capital costs, a state-of-the-art approach based design is expected to have smaller capital costs because suitable designs are rejected by the straightforward approach which are not rejected by a state-of-the-art model. This study also suggested, that the recirculation costs computed by a straightforward based model are overly optimistic, in case of a diffuser design. This could result in unforeseen costs for the operator.
Finally, the results obtained from this case study suggest that a state-of-the-art approach has limit added value when designing an open surface outfall but it is expected to be more when less advantageously scenarios are selected in the straightforward approach. Finally, the model results of a state-of-the-art based model approach are less sensitive to the models input and thus expected to be more reliable.
After two years of study, this thesis is my final project for the Master Civil Engineering and Management at the University of Twente. I am strongly interested in the mix between technology and management and this thesis is the perfect combinations of these two worlds. This master thesis has to goal to objectively investigate the differences between straightforward and state- of-the-art modeling methods when designing the intake and outfall configuration of a power plant. To this end, I set-up both a far and near field model, used a state-of-the-art coupling method and used the SBAM method to select ambient conditions. I was able to learn a lot about numerical modeling and I am all employees and students very grateful that they helped me in difficult times. Furthermore, working at Deltares gave me the opportunity to meet a lot of new interesting people with different backgrounds. I enjoyed to be able to participate and experience a famous consultancy company such as Deltares.
I would like to thank all the people who helped me with this thesis. First of all, Robin Morelissen for arranging this project, his expertise and time, the excellent help and keeping me positive in difficult times. I would like to thank Pieter Roos and Kathelijne Wijnberg for the good feedback and guidance. Thirdly, I would like to thank all the student at Deltares for the warm welcome, nice lunch breaks and the fun football tournament. Finally, I especially would like to thank Marnix for all his support, inexhaustible patience and interest.
Annet Both July 2016, Delft
Abstract iii
Preface and acknowledgments v
1 Introduction 1
2 Research framework 5
3 Case description and optimization framework 9
3.1 Case description . . . . 9
3.1.1 Study area . . . . 9
3.1.2 Power plant . . . . 11
3.1.3 Straightforward selection of ambient scenarios . . . . 12
3.2 Optimization framework . . . . 13
3.3 Costs assessment . . . . 16
4 Model development 19 4.1 Approach A: common consultancy practice . . . . 19
4.2 Approach B: state-of-the-art assesment . . . . 23
5 Results 27 5.1 General behavior of the intake and outfall system . . . . 27
5.1.1 Representative selection of ambient conditions (SBAM) . . . . 27
5.2 Comparison of the two approaches . . . . 39
5.3 Design choices . . . . 42
6 Discussion 47 6.1 Operational costs assessment . . . . 47
6.2 Intake and outfall configuration . . . . 48
6.3 Methods and model set-up . . . . 48
6.4 Representativeness of the case study . . . . 49
6.5 Addition model choice considerations . . . . 50
7 Conclusion and recommendations 51 7.1 Conclusions . . . . 51
7.2 Recommendations for further research . . . . 52 Appendix
A Results SBAM A-1
Introduction
The number of power plants is globally increasing and the capacity of existing power plants is extended. In coastal regions, some of these power plants use ambient water for cooling, which increases the temperature of the water used. In the so-called once-through cooling method, the sea-water is discharged back into the sea after the cooling process. Power plants may thus have a large influence on the ambient environment because the typical temperature increase is around 10◦C (Bleninger and Morelissen, 2015; World Nuclear Organisation, 2015). Furthermore, the discharge for newly built power plants can be up to hundreds of cubic meters per second (Morelissen et al., 2015). Therefore, environmental quality standards are set near the outfall location to protect to local natural environment. These standards have to be met to be allowed to build the power plant. High dilution rates are preferred in order to reduce the effects on the ambient environment as much as possible.
Regarding the plume’s behavior, we distinguish three zones; the near, intermediate and far field. The near field is dominated by initial momentum and the far field behavior is governed by ambient flow conditions. The intermediate field is the transition zone between these fields.
The effluent is relatively warm compared to the ambient water, which creates a positive buoyant plume that will rise to the surface. The outfall type, as well as other design parameters such as the discharge rate, has an influence on the plume’s behavior. The most common outfall type is a surface outfall because it is cheap and easy to build. However, the outfall plume is not quickly diluted. By using a diffuser outfall instead, at least a 5-10 times higher mixing efficiency can be reached (Jones et al., 2007). A diffuser is a long pipe with a (large) number of outfall ports.
However, the capital costs are much higher for a diffuser than for a surface port.
Additionally, the plume’s behavior has an indirect effect on the capital and operational cost of a power plant. Operational costs are related to the energy needed for operation, e.g. the amount of pumping needed, and the required cooling capacity, e.g. the intake temperature. If the intake temperature is higher, the power plant condenser will be less efficient regarding heat transfer. This results in a lower efficiency rate and consequently less power output and a lower revenue. Assessing the plume’s behavior can therefore help to make an optimal intake and outfall configuration in order to achieve the highest revenue. Computer models are an important tool for modeling the plume’s behavior. The design of a power plants configuration is most often based on these computer model predictions. The different plume behavior zones and ambient conditions are a challenge for modelers. Currently, separate models are available for the near field and far field, which are both dependent on the input values and assumptions made by the modeler. This makes it difficult for non-expert clients to judge the quality of the model outcomes.
Problem statement and objective The problem statement of this project is:
New modeling approaches have been developed to provide more accurate and comprehensive design information with the goal of a better design of the intake and outfall of a power plant. However, it is difficult for non-expert clients to identify the differences and possible benefits of the different
modeling approaches. Therefore, the added value of these improved modeling approaches should be objectively quantified to make the different model approaches distinguishable for non-expert clients.
Therefore, the research objective of this thesis is to objectively investigate the added value of state-of-the-art recirculation modeling compared to a typical straightforward modeling approach for the optimization of the intake and outfall configuration of a power plant.
The main research question of this thesis is:
Could the usage of a state-of-the-art recirculation modeling approach result in a better intake and outfall design, compared to a straightforward model approach?
Theoretical background and model definition
A thermal power plant produces electrical energy in 4 steps, based on the so-called Rankine cycle. First, the water is converted into steam inside a boiler, for example by burning fossil sources or by a nuclear process (Turchi et al., 2010). Hereafter, the steam is transported into a turbine, where a shaft is set to motion linked to a generator which produces electric energy (Mohsen, 2004). The steam leaves the turbine and is condensed into water (Mohsen, 2004).
As the temperature difference between the condenser water and the external water decreases, the heat exchange also decreases resulting in a decreasing power plant efficiency. Therefore the external water temperature is preferred to be as low as possible. Only a few studies investigated the relation between intake temperature and the efficiency of a power plant. However, Tramel (2000) describes the relation between the heat rate of the condenser compared to the intake tem- perature. The heat rate is a unit that describes the amount of energy input needed to produce a certain output, which can be understood as the reverse of the efficiency of a power plant.
There are three commonly used numerical model approaches in the near field (Palomar et al., 2012). The integral model approach is a detailed, physically correct and typically used model approach. The differential equations of transport and water motion cross-sectionally integrated in order to make them easier to solve. Commonly used integral models are CORMIX (Jirka et al., 1996), VISJET (Cheung et al., 2000) or Visual Plumes (Frick, 2004). They all have their own approach and should be used with caution to avoid unreliable model results (Palomar et al., 2012;
Schreiner et al., 2002). For quick understanding of a complex case, a dimensional analyis model approach could be used (Palomar et al., 2012). An example of this approach is the classification of the plume behavior by Jirka et al. (1991). Finally, a Computational Fluid Dynamics (CFD) model approach is the most comprehensive and accurate model approach. It computes all flow characteristics at defined points in a grid, but the computational time is longer. Therefore, this model approach is less commonly used in the near field (Palomar et al., 2012; Bleninger, 2006).
However, it is becoming more used.
In the far field, a numerical model is recommended because the results are accurate and the grid sizes are larger resulting in acceptable computation time. Examples of this type of hydrostatic 3D-models are Delft3D, MIKE and EFDC. These models can be classified in terms of their numerical schemes (Bleninger, 2006).
A simple solution to model the entire plume trajectory is to use the far field model also in the near field. This has the disadvantage of the results becoming very sensitive to the chosen grid sizes, especially the dilution and plume width (Bleninger and Morelissen, 2015). These kinds of methods are often the foundation of the design of an intake and outfall configuration. In this MSc project, this approach will be called Approach A.
A state-of-the-art approach is coupling the near and far field. Each model is then used to predict the characteristics in their own zone but the models exchange relevant information to predict the entire plume trajectory. Blumberg et al. (1996) and Zhang (1995) were the first investigating the coupling of the near and far field. They found that the modeled trap height, i.e. the plume’s rise, and initial dilution were similar for both near and far field models. Hereafter, Zhang and Adams (1999) introduced several coupling methods based on the predicted trap height of the near field model. Several later studies showed that coupling the near- and far-field models give accurate outcomes (Kim et al., 2002; Suh, 2006; Morelissen et al., 2013; Nekouee et al., 2015). A state-of-the-art coupling approach is the Distributed Entrainment Sinks Approach (DESA) (Choi and Lee, 2007). In this method entrainment sinks are incorporated to preserve the mass balance. It is considered to improve the physical representation of the plume’s behavior (Choi and Lee, 2007; Bleninger and Morelissen, 2015).
Another state-of-the-art model approach is the so-called Scenario Based Adaptive Modelling (SBAM) (de Fockert et al., 2011; Verbruggen et al., 2014). This is a comprehensive approach to deal with varying ambient conditions. It involves incorporating the most common plume trajectories instead of the most common ambient conditions. Both de Fockert et al. (2011) and Verbruggen et al. (2014) state that this method results in more reliable and representative model outcomes than the traditional approach.
For this research, the State-of-the-art approach is a model that includes both a near and far field model. These are coupled with the DESA method in combination with ambient scenar- ios selected based on the SBAM method. This model approach will be referred to as Approach B.
Outline of methodology
Because of the wide range of variation in the above stated problem, we tackle this problem with a case study. This case was carefully chosen in order to capture the most important and interesting parameters. Two models were set up, the first using a straightforward approach (Approach A) and the second using state-of-the-art model approaches (Approach B). Approach A is a 3D far field model. This model is extended by a dynamically coupled near field model and the SBAM method to form Approach B. Based on these two models, two offshore intake and outfall configuration designs were made. In order to assure an objective design process, a design framework was set up beforehand. This framework includes 18 predefined designs and fixed criteria to select the ’best’ design option. The differences between the two model approaches were estimated in terms of operational costs, based on the known relation between the heat rate and the intake temperature in the power plants condenser. Furthermore, the offshore capital costs were considered. The research choices that were made to create a feasible study are described in Chapter 2. The methodology and model set-up are described in more details in Chapter 3 and 4 respectively.
Relevance and innovation
This project will bring insight regarding the need for comprehensive models of plume behavior.
This study is one of the first projects to link model approaches to operational costs of a power plant.
As well, this project is useful for both consultancy agencies and their non-expert clients, since it will give consultancy agencies an extra tool to show the quality of a model and it helps them to get insight in the effectiveness and need for comprehensive modeling.
This research could also be interesting for the desalination sector, where currently the same model approaches for the assessment of the plume behavior are used. Desalination is the pro- duction of fresh water by removing the salt from salty water, for example from seas or brackish
estuaries. Currently the same model approaches for the plume behavior are used in this study field. Therefore, the outcomes of this study might also be valid for modeling a brine plume.
Reading guide
First, Chapter 2 will describe the research requirement and develops a rough research framework.
Chapter 3 will give all used method in detail. In this chapter, the characteristics of the study area will be described followed a description of the optimization framework. This chapter ends with a section on the costs assessment and model description. Hereafter, Chapter 4 will describe the model set-up of both approaches. Chapter 5 presents all results. First by comparing the two approaches and then comparing the design choice. All findings will be discussed in Chapter 6.
Finally, this reports ends with the conclusions and recommendations in Chapter 7.
Research framework
The research objective described in the previous chapter covers a wide range of processes. Fur- thermore, the objectiveness of this study should be secured. Therefore, the used methods in this study should be chosen such that these requirements are met. This chapter describes the conditions which the methods should fulfill.
Case study
The introduced problem was investigated with a case study because the research objective set in Chapter 1 involves a wide range of variation in parameters. The disadvantage of a case study is that it is not all encompassing. Thus, the case area should be carefully chosen such that it is representative. Therefore, the case area should be a typical and complex area. It should be forced by several relevant and typical ambient processes. Furthermore, none of these processes should be highly dominating the system. The details of the chosen case area are given in Section 3.1.
The investigated power plant should have a typical temperature increase. The discharge should be relatively large because the effect on smaller discharges can be estimated based on this.
Optimization and selection framework
In this research, a ’best’ design should be recommended to a fictions client. This involves an optimization process. However, the research objective asked for an objective and fair optimiza- tion process. Therefore, an optimization framework was set up. This framework should be feasible considering the amount of simulations because of the expected long simulation times.
Furthermore, this framework should be able to handle the large amount of correlated variables.
Therefore, 18 possible design options were set-up in advance of which the ’best’ option will be chosen. The variables that are expected to cause the largest differences are the intake/outfall type and their location. It should be noted that this framework will not result in the optimal design, because the designing process is more iterative and more parameters are involved when a site specific design is made. However, such a precise optimization is not the goal of this project.
Additionally, the objectivity of this study is guaranteed by the use of a selection scheme for selecting the ’best’ design option. The scheme was set up based on common requirement from the power plant developers. This in order to simulate the design processes the most realistic.
Section 3.2 will give the site specific characteristics of this framework and the used selection framework.
Cost assessment
In this study, we use the capital costs (CAPEX) and operational costs (OPEX) of a power plant to investigate the added value of state-of-the-art modeling. The scope of the research is to create a design of the intake and outfall configuration only. Therefore, only the offshore costs are considered in the cost assessment.
The operational costs of the intake and outfall configuration of a power plant consist of many elements such as; electricity costs for the pumps, head loss costs, cleaning of the intake screens and recirculation costs (DNVL-GL, 2015). In this study, only the recirculation costs are taken into account because this is strongly related to the modeled intake temperature. Furthermore, the relation between intake temperature and recirucaltion costs is not yet investigated and is therefore a new interesting element. Section 3.3 describes the costs assessment in detail.
Model requirements
Many models are available that assess the plume behavior. The accuracy and complexity are different per model approach. Figure 2.1 shows an overview of the available models. It is very important to note that the accuracy is highly depended on site specific characteristics.
Figure 2.1: Very rough comparison of the available models to assess the plumes behavior. Please note that the accuracy is highly dependent on the ambient characteristics.
The most simple approach to model the plume behavior are Rapid Assessment Tools such as the Length scale analysis and flow classification or the use of Empirical Dilution Equations (Bleninger and Morelissen, 2015). For example, initial fluxes and ambient conditions are used to classify the plume behavior in the Length scale analysis. These tools can only incorporate simplified ambient conditions and are therefore only used for a quick initial assessment to identify unproblematic discharges. These approaches are not complex enough to be used for designing the intake and outfall configuration.
Most often, far field models are used to design the intake and outfall configuration. However, the accuracy of the far field model is dependent on the model choices. For example, the accuracy is lower if a 2D model is created instead of a 3D model. A 2D model is not capable to assess the plume behavior accurately because it is expected that the large water depth in the grid point dilutes the plume quickly. The disadvantage of a 3D far field model is that it is not able to assess the near field accurately. However, this model approach is a common consulting standard.
Another option is to use a near field model only, but these models are less accurate in the far field. Therefore, they are less suitable
The most complex option is to couple a near and far field model. This can be carried out using a one-way or two-way coupling method (Morelissen et al., 2015). A one-way method includes a fixed diluted source, predicted by the near field model, as input for the far field model. Two-way coupling, also referred to as dynamic coupling, is a method that incorporates the interaction between the near and far field processes by updating the ambient conditions in a sufficiently small time interval (Zhao et al., 2011). Morelissen et al. (2015) showed, in a case study of a large
outfall system of a power plant, that dynamic coupling represented the physical processes better than an one-way coupling method.
To be able to give a fair and reliable answer to the research question, the state-of-the-art model (Approach B) should be compared with a model that is a common consulting standard (Approach A) and is able to reasonably accurately assess the behavior of the plume. Furthermore, Approach A has the purpose to be a simple and quick model that is able to capture the thermal plume behavior and ambient conditions influencing this process. Therefore, Approach A will be a 3D far field model. Approach A is able to incorporate the variation of all ambient conditions in the case area and the differences in temperature and salinity. Furthermore, the model should have simulation duration of maximum 24 hours.
Approach B should be able to capture the same processes as Approach A. However, this model approach has to assess the physical process more accurately. Therefore, the far field model of Approach A is extended by state-of-the-art model approaches to form model Approach B. The state-of-the-art model approaches we use in this study will be the SBAM scenario selection method and the far field model is dynamically coupled with a near field model. The main difference between the scenario methods is thus that the scenarios of Approach A are chosen before modeling starts. Approach B includes SBAM which finds the best ambient conditions during modeling by uses an iterative approach.
Approach A: a 3D far field model forced by ambient scenarios that are selected based on common weather conditions, see details in Section 4.1 and 3.1.3 respectively.
Approach B: the far field model of Approach A is dynamically coupled with a near field model, see Section 4.2. In addition, the SBAM method is used to select the ambient scenario, see Section 5.1.1.
Case description and optimization framework
This chapter describes the methods used in this study. First, the description of the study area is given. Hereafter, the selection framework is elaborated in section 3.2. Finally, the cost assessment is presented in section 3.3.
3.1 Case description
In order to answer the research questions, first a fictitious power plant and study area were chosen.
Such study case represents a typical assignment for Deltares, as already stated in Chapter 2. The fictitious study area will be described first in Section 3.1.1, followed by the characteristics of the power plant in Section 3.1.2.
3.1.1 Study area
The selected case area is Vung Ang in the Northern half of Vietnam. This area fulfills all requirements set in the previous chapter. The Vung Ang area contains several relevant and typical ambient processes in this field of study. Some of these processes are: a river discharge near the power plant, a residual current and varying wind. None of these processes is highly dominating the system. This case was based on a typical assignment that Deltares carried out.
An overview of the shoreline and an overview of the influencing processes are given in Figure 3.1, and details of the ambient processes will be discussed now. Detailed information is required in order to be able to perform a SBAM based data selection. Furthermore, this is required to keep this study realistic.
Wind Due to the positive buoyancy of the plume, the wind can have an influence in the far field.
Changing wind conditions could alter the plume’s direction. Furthermore, the cooling of the plume can intensified by a strong cool wind. For this case study, typical wind conditions were extracted from the CFSR database. A typical wind rose from the wind data between 2000-2010 is presented in Figure 3.2. The wind direction is strongly varying in the area with an average wind speed around 3m/s. Storms seem to occur mostly from the North. Another remarkable feature of this area is that in some months the wind direction is varying during the day. A good example of this is March, as can be seen in Figure 3.3. This variation was found in March till May and in August and September. A hourly changing wind direction could have an impact on the plume dispersion.
Tide This study area is forced by an semidiurnal tide. The tidal range in the area is up to 1.6 meters. It is expected that the variation in the tide during the year has a relatively small impact on the plume’s behavior compared to the other processes. However, the variation in the spring-neap circle will have an influence because of the changing water levels.
Figure 3.1: Overview of study area with ambient conditions, modified from de Fockert et al.
(2011).
Figure 3.2: Wind Rose for this case study based on data 2000-2010 (CFSR, 2016).
Figure 3.3: Hourly variation in March, produced from data between 1999-2009 (CFSR, 2016).
River Tropical regions generally have a dry and a wet season, hence the river discharge of this case can be classified into a dry, January till July, and a wet, August till December, season discharge. In Table 3.2, an overview is given of the river discharge of this fictitious river, based on Deltares (2010). It was estimated that in an extremely wet season the maximum discharge can be up to 170 m3/s.
Water depth The case area consist of a gentle and constant slope. No large irregularity are noticeably besides the land abutment in the middle of the case area. The depth profile is given in Figure 4.3.
Residual Current Manh and Yanagi (2000) showed that residual current in the Gulf of Tongk- ing is induced by wind. The case area in this study is located in this gulf, and therefore the residual current found by Manh and Yanagi (2000) are used in this study. The range of the residual current can be found in Table 3.1. The residual current in related to the seasonal variation in the wind field.
Temperature Table 3.2 shows some typical mean air and sea temperatures for a tropical region.
The difference between the air and sea temperature is small or even equal.
3.1.2 Power plant
The intake and outfall configuration for this study will be carried out for a ficitous power plant with the following characteristics:
• A constant intake and outfall discharge of 50 m3/s.
• A constant relative outfall temperature of ∆T = 8◦C compared with the intake tempera- ture. This is a typical value for power plants.
• The power plants consists of four units each producing 150 MW, producing 600MW in total.
Table 3.1: Residual current per month according to (Deltares, 2010) and hourly wind variation.
Month
Residual current velocity
[cm3/s]
Residual current direction Hourly wind variation
January 0/5 SE Persistent
February 0/5 SE Persistent
March 0/5 SE Variable
April 0/5 SE Variable
May 0/5 SE Variable
June 0/3 SE/NW Persistent
July 0/3 SE/NW Persistent
August 0/3 SE/NW Variable
September 0/3 SE/NW Variable
October 0/5 SE Persistent
November 0/5 SE Persistent
December 0/5 SE Persistent
Table 3.2: Typical river discharge, tropical air and sea temperatures used for this case study
Month
River
discharge [m3/s]
(Deltares, 2010)
Mean air temperature [◦C]
(The Weather Company, 2016)
Mean sea temperature [◦C]
(Deltares, 2010)
January 6 19 19
February 1 21 20
March 0 24 22
April 0 26 25
May 0 32 28
June 0 32 29
July 0 29 29
August 18 30 29
September 76 29 29
October 118 26 27
November 70 25 24
December 36 18 21
3.1.3 Straightforward selection of ambient scenarios
Based on the characteristics described in the previous chapter, a quick and commonly used method to select the ambient scenarios can be carried out. In this method, ambient scenarios are selected based on on non-extreme weather conditions. This is a common practice and will be used in Approach A. Therefore, the scenarios for Approach A are based on the two main seasons in a tropical region: the wet and the dry season. Looking at the river discharge, a clear distinction can be made; Augustus till January is the wet season and February till July is the
dry season, see also Table 3.2. For the dry season, a river discharge of 0 m3/s is the typical event, whereas the wet season will be represented by a discharge of 70 m3/s.
Based on these two seasons, representative wind conditions are selected. From Table 3.1 it can be seen that within the dry and wet season the wind direction can highly differ. Therefore, different wind conditions have to be picked for both seasons to resemble scenarios for both seasons. The average wind speed is constant during the day in all months.
Based on this analysis, a varying and persistent wind direction was selected for each season.
January represents a persistent wind condition in the wet season, the persistent wind condition for the dry season will be represented by the month July. A varying wind will be represented by the month September for the Wet season and March for the dry season. From the data, a representative year was chosen, this was based on the monthly variation. The applied water temperature differs per ambient scenario, see Table 3.2. Therefore, the corresponding averaged ambient temperatures for each season was imposed to assure the most realistic scenarios. The chosen ambient scenarios for Approach A can be found in Table 3.3.
Table 3.3: Ambient scenarios for Approach A
Scenario Season Discharge
[m3/s] Wind condition Ambient temperature [◦C]
1 Dry 0 Persistent (July 2008) 29
2 Dry 0 Variable (March 2009) 22
3 Wet 70 Persistent (January 2007) 19
4 Wet 70 Variable (September 2006) 29
3.2 Optimization framework
Chapter 2 described that 18 possible design options were set-up in advance of which the ’best’
option will be chosen. This section describes the site specific characteristics of these 18 design options. Parameters such as the diffuser orientation and angles are chosen to be constant because these can not be modeled in Approach A and are therefore less important. Thereafter, the site specific selection framework is presented. The following variables are different in each design:
Intake type Two intake types are combined for nine different outfall types. These intake designs are a channel intake and a submerged intake. An example of the intake types is presented in Figure 3.4 and 3.5. In general, it is expected that intake temperatures are lower when using a submerged intake because the water temperature is lower in the bottom layers.
However, these are more expensive. The submerged intake will be 300 meter away from the coast and at 6.3 meter depth.
Outfall type We test two different outfall types; the diffuser and the surface outfall. Examples of the layout of these outfall types are given in Figure 3.6 and 3.7. These two are very common but differ in mixing capacity and capital costs. A diffuser is more expensive to build, however it has a high mixing capacity. The surface outfall is cheap, but has low mixing capacity. For simplicity, the diffuser and outfall dimensions are fixed. The typical diffuser used in this assessment has a length of 100 meter, a port diameter of 2 meter, and 6 openings. The vertical angel of the nozzles is 15◦above the horizontal plane. All nozzles are orientated along the diffuser. A top view and side view of this diffuser is shown in Figure 3.8. The diffuser is laying 30◦ from North and submerged at the seabed.
Figure 3.4: Example of a channel intake (Canadianpond, 2016)
Figure 3.5: Example of submerged intake system (Bleninger and Jirka, 2010)
Figure 3.6: Example of a diffuser outfall pipe (Doneker, 2014)
Figure 3.7: Example of surface port outfall system (PembangkitListrik, 2015)
Figure 3.8: Top view and sideview of the designed diffuser.
Outfall location Because surface outfalls are always located at the shoreline, the location of the surface outfall along the coast is variable whereas this x-location at the coast is fixed for the diffuser outfalls. For submerged outfalls, the water depth has an influence on the initial mixing. Therefore, the diffuser design is tested for different location away from the coast. This results in 6 diffusers at a different distance from the coast between 1250m and 2400m with increments of 230m. Furthermore, 3 surface outfall at the coast line with increments of 250m will be tested. Different location in x-direction is less interesting because the differences are expected to be smaller. Figure 3.9 gives an overview of the different locations of the outfalls. The nearest and furthest diffuser from the coast are located at 9.6 and 11.9 meter depth respectively. An overview of the specification per design is given in Table 3.4.
Figure 3.9: Overview of used locations of the different designs in the optimization process.
Selection criteria
From the 18 design options, one design will be selected as the one that is recommended to the fictitious client. A fixed selection scheme was produced to keep an objective view. This scheme was set up based on common requirement from the power plant developers. This results in the following scheme:
1. All designs that do not meet the environmental criteria are discarded. The World Bank Group (1998) recommends the following environmental criteria when no local criteria are set:
”The effluent should result in a temperature increase of no more than 3◦C at the edge of the zone where initial mixing and dilution take place. Where the zone is not defined, use 100 meters from the point of discharge when there are no sensitive aquatic ecosystems within this distance.”
In conclusion, the maximum excess temperature is not allowed to be more than 3◦C at more than 100 meters from the discharge point at 1 meter below the surface.
2. All designs with a maximum exceedance temperature higher than 1◦C at the intake, are omitted. This is a requirement that is often demanded by the power plants owners.
3. From the remaining designs, the most cost-efficient option is selected as the final design.
In practice, this most often based on the capital cost only. However, this might not lead to the cheapest option in the long term. Therefore, it will be investigated whether a different design is selected when also the operational cost are considered in this step. Operational cost are in this study only estimated based on the recirculation costs. This results in two selection methods; one based on the common practice and one more sophisticated.
a) Only the capital costs are included in the calculation, operational costs are excluded.
b) Both the capital and operational costs are considered. The operational costs will be estimated based on the temperature at the intake according to the model that was used here. Furthermore, a service life of at least 25 years is assumed (International Energy Agency, 2005).
Table 3.4: Specifaction per design
Design # Intake Outfall Distance Distance
from coast [m] from base[m]
1 Submerged Diffuser 1250
2 Submerged Diffuser 1480
3 Submerged Diffuser 1710
4 Submerged Diffuser 1940
5 Submerged Diffuser 2170
6 Submerged Diffuser 2400
7 Submerged Surface -250
8 Submerged Surface 0
9 Submerged Surface 250
10 Surface Diffuser 1250
11 Surface Diffuser 1480
12 Surface Diffuser 1710
13 Surface Diffuser 1940
14 Surface Diffuser 2170
15 Surface Diffuser 2400
16 Surface Surface -250
17 Surface Surface 0
18 Surface Surface 250
3.3 Costs assessment
For each design in the optimization process both the offshore capital and recirculation costs were estimated, as was described in Chapter 2. The following steps were taken considering the estimation of the recirculation costs:
The relationship between the intake temperature and the condenser can be described by the net unit heat rate (q), also called the the reverse of the power plants efficiency (η),
η = q−1 (3.1)
where the efficiency (η) is a percentage and the Net Unit Heat Rate (q) is in kW h/kW h.
In other words, the net unit heat rate represents the amount of energy needed to produce one
kilowatthour of electricity. Thus, a power plant with a lower heat rate is able to produce more energy with he same mount of energy than a power plant with a higher heat rate (International Energy Agency, 2015). This is the initial concept of the assessment of the recirculation costs based. Knowing the Gross load (P ) of the power plant and the power plants efficiency η, the Net power generation (Pg) was computed as follows,
Pg= P η. (3.2)
In this case study, the Gross load (P ) are 4 units of 150 MW. The Net power generation is therefore also in MW. To compute the yearly power output (Ey), the Net power generation is multiplied with the production time (tp) of the power plant. It was assumed that the power plant operates all year through at full power,
Ey= Pgtp. (3.3)
The potential Revenue ($/year) of a power plant can then be computed as:
Revenue = 84Ey (3.4)
A selling price for electricity of 84 $/M W h was used in this formula. This is an estimation of the global selling price according to Department of Energy and Climate Change (2016).
Finally, the relation between the amount of recirculation and the additional operating costs, for the power plant as described in Section 3.1.2, can be assessed using the formulas above and the relation between intake temperature and the net unit heat rate. The net unit heat rate for a 150 MW power unit is 2370 kCal/kW h at an intake temperature of 22 ◦C (Tramel, 2000).
The power plant is more efficient at an intake temperature of 18◦C, with a net unit heat rate of 2360 kCal/kW h. This net unit heat rate for an intake temperature of 22 and 18 is equal to an efficiency of 36.280% and 36.434% respectively. Due to a lack of more detailed information, the relation between these two parameters is assumed to be linear and valid for higher intake temperatures. The recirculation costs are estimated as the additional costs for an power plant compared to an ideal case with zero recirculation. The relation between intake temperature T and the Recirculation Costs (RC) in dollars can be described with the following formula:
RC = 84P ∆ηtpT (3.5)
RC = 170.000T (3.6)
A yearly averaged recirculation temperature of 0.5◦C would result in yearly costs of $85.000.
In 25 years, this rises to an additional cost of 2 million dollars compared to zero recirculation.
In the selection phase of this study, Section 5.3, the costs will be estimated with the temper- ature found by the model for which the design is selected. However, during the final evaluation between the two selected designs by the different approaches we consider the SotA-approach as the best estimation of the system because no measurements to validate are available. Studies showed that state-of-the-art approaches result in more accurate system estimation (Bleninger and Morelissen, 2015; de Fockert et al., 2011; Morelissen et al., 2015). Therefore, this was also assumed in this study.
To compute the costs per year for each design, the weighted average temperature will be multiplied with Equation 3.6. An inflation rate (i) of 3% is used to computed the total recircu- lation costs in the lifetime of the plant (San Diego County Water Authority, 2009). The total recirucaltion costs will be computed as:
RClif etime= RCyearly
1 − (1 + i)n
−i (3.7)
where n is the expected lifetime of 25 years of the power plant.
The offshore capital costs include the costs for the diffuser or surface outfall, the submerged intake or surface intake and the required pipeline length. The price per unit can be found in Table 3.5 and is obtained from feasibility studies.
Table 3.5: Price per unit for the offshore part of the intake and outfall configuration.
Price per unit
Diffuser1 $ 1.500.000
Surface outfall2 $ 1.000.000 Submerged intake2 $ 3.000.000 Surface intake2 $ 1.000.000 Pipeline [per meter]3 $ 1.500
1Dannenbaum Engineering Corporation (2004)
2Watereuse Association (2012)
3Dannenbaum Engineering Corporation (2004); San Diego County Water Authority (2009)
Model development
This chapter will present detailed information on the used models. First, the far-field model of Approach A will be described. Hereafter, the used extensions on this far field model to create Approach B is given in Section 4.2.
4.1 Approach A: common consultancy practice
As chosen in Chapter 2, Approach A is a 3D far field model. This section will will describe all settings in the model that are needed to meet the requirements set in Chapter 2. The model was built in the hydrostatic Delft3D software. Delft3D is an open source software package developed by Deltares (Lesser et al., 2004). It is a 2D (depth-averaged) or 3D model that can simulate unsteady flow and transport phenomena including density differences. The hydrodynamic part is computed with the use of horizontal equation of motion and the continuity equation. An ad- vection/dispersion equation is incorporated to describe the transport phenomena. The resulting horizontal density differences are then included in the hydrodynamic part. Delft3D-FLOW was designed to describe phenomena where the vertical length and time scales are much smaller than the horizontal scales. This makes it useful for the far field modeling of the plume. Delft3D-FLOW uses the so-called “shallow water assumption” to simplify the vertical momentum equation. This assumption is valid when the vertical flow accelerations are negligible compared with gravity.
In the near-field, this assumption in not valid because the buoyancy effects result in vertical acceleration. (Lesser et al., 2004; Morelissen et al., 2013; Bleninger and Morelissen, 2015)
The model incorporates the influence of wind.
• The temperature in vertical and horizontal direction of the ambient sea and river is uniform.
• The temperature of the river is equal to the ambient sea water.
• The salinity in the sea and river is uniform in horizontal and vertical direction.
• It is expected that the bed roughness has small influences on the model results and therefore it was chosen to keep the bed roughness constant in the whole domain. Furthermore, change in the system due to a changed bed roughness is not within the scope of this project.
• Changes in bed morphology over time do not influence the plume behavior. Therefore, this process can be neglected in the model.
Computational grid
A domain area of 20×30 km2 was created to model the plume behavior, see Figure 4.1. The grid size near the outer boundaries are up to 500×300 m2. To be able to capture the small scale plume behavior near the outfall, the grid size is decreasing towards the area of interest with the smallest grid size being 40×40 m2. A detailed presentation of this part of the grid is presented in Figure 4.2. In this figure, also the modeled river is clearly visible. The river length is chosen to be as 1000 m long, such that the model captures the tidal influence on the river.
Figure 4.1: Overall grid.
Figure 4.2: Refined grid in the area of interest.
The bathymetry was created by first extracting depth points from the Delft Dashboard soft- ware using the GEBCO 08 database as source. Next, these depth points were filtered, only negative values were kept in order to exclude dry areas inside the grid. These missing values were then interpolated with the surrounding values. The resulting depth contours of the whole domain can be found in Figure 4.3 and the depth contours of the detailed area is shown in Figure 4.4.
Vertically, the model includes 10 layers, with the purpose of accurately accounting for the stratification of the thermal plume. To be able to achieve an increasing vertical resolution towards the more shallow regions, a σ-layer setting was used to get a constant number of layers. Each vertical layers was set to represent 10% of the total local depth.
Figure 4.3: Depth profile of the overall grid.
Figure 4.4: Depth profile of refined area of the grid.
Figure 4.5: Model with and without residual current, example location in the North-West.
Time step and modeled period
Based on the Courant stability criteria, a time step of 0.25 min was used for all scenarios. No significant differences were found when comparing the model results of a reference run with a model run using a time step of 0.1 min. Therefore, the time step of 0.25 min was certified as appropriate for this problem whereas a larger time step, 0.5 min, did result in major differences and was not suitable. One month was modeled per simulation. From some test simulations it was concluded that the first week was required as spin-up time. The last three weeks were used to analyze the model results. The simulation time of the model is approximately 22 hours, run parallel on a 3 core cluster.
Tidal boundary conditions
The eastern and western open boundaries are forced with current data. On the northern bound- ary, the water level is prescribed. The input of these boundaries are based on a larger computation grid (300×400 km). This process is also called ’nesting’. This model was built in order to make more stable conditions near the edges of the computational grid. This large overall 2D grid had a resolution of 1000m. The tide forcing data on the boundary were obtained via Delft Dashboard from the TOPEX7.2 Global Inverse Tide model database. The large overall model was then run with the boundaries of the model grid boundaries as observation points. Hereafter, this modeled was nested and the boundaries were created. This means that the values imposed at the boundaries are based on a time series.
To reduce the amount of reflection in the boundaries of the overall, a reflection parameter α of 1500 was applied at the current boundaries, and 2500 at the northern water level boundary. The salinity and temperature are constant in time and depth. However, for the different scenarios, represented by a certain month, the applied ambient temperature differs. These different ambient temperatures are shown in Table 3.2.
Residual current
The residual current in Approach A is taken to be constant because the variation in the parameter is relatively small, see Table 3.1. A residual current of 3 cm/s SE was chosen as typical event. This phenomenon is included in the model by superposition of the velocity on the current boundaries, the result can be found in Figure 4.5. This figure shows the model velocity with and without the residual current in the North-West of the modeled area.
River discharge
The river was modeled as a boundary condition, a total discharge of boundary was used to include
the extra discharge in the model. The salinity of the water was set to be 10 ppt, a representative value for brackish water.
Intake and outfall
The intake and outfall configuration was model by using the in-out discharge setting in Delft3D.
This setting makes it possible to subtract a certain discharge and add this back into to the system at another location with an increase in temperature.
Ambient temperature and wind conditions
The ambient sea temperature is modeled as a constant. Varying temperature in one simulation is beyond the scope of this research. The wind speed and ambient air temperature are used to compute the heat exchange between the sea water and the ambient environment, also called the excess temperature model in Delft3D. The air and sea water are modeled with the same temperature for two reasons. First, it is undesirable that the sea water cools down as a result of lower ambient air temperatures, this will affect the ambient sea temperature near the outfall and is therefore not a good representation of the plumes ambient environment. Secondly, the differences are small as can be seen in Table 3.2.
Ambient scenarios
The ambient scenarios for Approach A are described in Section 3.1.3.
4.2 Approach B: state-of-the-art assesment
The aim of the state-of-the-art based model is to use up-to-date model techniques to assess the plume behavior as accurately as possible. Therefore, Approach B contains a near field and far field model. This section will describe the implementation of the near field and the applied coupling method. The far-field model that was used in Approach A, see Section 4.1, was also incorporated in Approach B.
Near field model
The Cornell Mixing Zone Expert System (CORMIX) software was selected to assess the near field plume behavior. This software was developed to predict the dilution characteristics and geometry of the initial mixing zone of an outfall plume (Doneker and Jirka, 2007). CORMIX computes the near field behavior of the plume, based on the characteristics of the outfall and discharge, such as nozzle orientation and initial density. Furthermore, CORMIX uses the ambient conditions such as the ambient velocity, density differences and the water depth to predict the near field behavior. With this input, momentum and buoyancy fluxes are used to develop length scales. The magnitudes of these length scale are used to classify the plume in different zones, as described by Jirka et al. (1991). Furthermore, CORMIX uses an additional integral approach (CorJET) to assess the intermediate zone of the plume. The output of CORMIX includes the direction, thickness and width of the plume as well as the dilution rates.
This software proved to predict the near field plume behavior accurately (Palomar et al., 2012; Doneker, 2014). The CORMIX software is suitable to model a multiport and positive buoyant effluent (Morelissen et al., 2015). Therefore, this method will only be applied on the diffuser outfall designs and not on the open surface outfalls.
Coupling method
Every 60 modeled minutes, the near and far field model were coupled. This is a small enough time step to capture the variability in the system but it is large enough to not increase the simulation time with more than half a day.
The two-way coupling was performed with the COupled SUbgrid MOdel (COSUMO) ap- proach. This is an interface program for coupling Delft3D and CORMIX, see Figure 4.6 where the steps of the two-way coupling method are shown. The main advantage of this approach is that COSUMO translates and adapts the Delft3D output as CORMIX input and vice versa.
Pre- and post processing functions in COSUMO, make it possible for the user to define at which location the models are coupled. The coupling can be based on plume characteristics such as the width and dilutions rate or based on spatial characteristics such as the water depth or distance from the discharge point. Based on expert judgment, the couple location in this study is located once the plume fills less than half the water depth. At this location, the plume is restratified and most of the near field trajectory is formed. After trial simulations, it was found that for some time steps, the ambient velocity was very low. This is probably due to a change in tide direction. As a result, the recommended coupling location was far outside the near field zone, which is undesirable. Therefore, a post processing function was created where the maximum coupling location is 500 meter from the outfall. This ensures that CORMIX only predicts initial mixing in the zone where it proven to give valid answers. Hereafter, the simulation is taken over by the far field model.
Furthermore, the DESA method was applied in this study because it showed to improve the physical representation of the plume’s behavior (Choi and Lee, 2007; Bleninger and Morelissen, 2015). In this method, entrainment sinks are incorporated to preserve the mass balance. This was incorporated in the post-processing function. The sinks were not placed on the center line as suggested by Choi and Lee (2007) but by a state-of-the-art ’spiral sink’ method, a method where the entrainment sinks are placed around the outer perimiter of the plume. This is in order to achieve more accurate model results.
Ambient scenarios
The ambient scenarios for Approach B will be described in Section 5.1.1.
Figure 4.6: Flow chart showing the CORMIX-COSUMO-Delft3D interaction for the two-way coupling method.
